Long-term trends in temperature, precipitation and snow conditions in Northern Russia

Transcription

1 no. 9/2010 Climate Long-term trends in temperature, precipitation and snow conditions in Northern Russia Pavel N. Svyashchennikov and Eirik J. Førland

2

3 report Title Long-term trends in temperature, precipitation and snow conditions in Northern Russia Section Climate Author(s) Pavel N. Svyashchennikov 1 and Eirik J. Førland 2 Date //2010 Report no. /2010 Classification Free Restricted ISSN 1). State Institution Arctic and Antarctic Research Institute (AARI), St. Petersburg, Russia 2). Norwegian Meteorological Institute (met.no), Oslo, Norway e-issn Client(s) Client s reference This study is part of a collaboration between met.no and AARI on the Norwegian Research Council projects «EALÁT- Research: Reindeer herders vulnerability network study: Reindeer pastoralism in a changing climate» and CAVIAR: Community Adaptation and Vulnerability in Arctic Regions Abstract Long-term variability in temperature, precipitation and snow conditions has been studied for 18 localities in northern Russia. An overview is given of station metadata and of methodology to study long-term trends and shifts in various climate series. For temperature no statistically significant trends ( ) were found in the annual average series for Kola Peninsula, Nenets AA and Yamalo-Nenets AA, Komi republic, while there are significant positive trends in the data for Sakha republic and Chukci AA. At most stations there are a significant warming after A negative trend ( ) was found in number of days with very low temperatures (daily mean temperature <-30 C). In the Chukci district there was an increase in number of days with mean temperature above 0 and +10 C. Also for the Kola Peninsula there has been an increase in number of days with mean temperatures > 0 C. For precipitation significant positive trends ( ) were found for all seasons except summer,- with strongest trends during winter and spring. The strongest trends were found for the Lovozero station at the Kola Peninsula. The precipitation trends are however influenced by inhomogeneities caused mainly by changes in precipitation gauges during the period For duration of snow cover two stations (Mare-Sale and Gigansk) had significant trends. At both stations the trends were positive, i.e. increased length of the period with continuous snow cover. The highest snow depths in northern Russia are usually observed in April. For 4 stations significant positive trends were found in snow depths for April;- the strongest increase is observed at Tarko-Sale and Ust Cilma. The reason for the increase in snow duration and maximum snow depths despite increasing temperatures is probably increased amounts of solid precipitation during the cold season. Keywords Temperature, precipitation, snow, trends, Northern Russia Disciplinary signature Responsible signature Øyvind Nordli Eirik J. Førland Postal address P.O.Box 43, Blindern NO-0313 OSLO Norway Office Niels Henrik Abelsvei 40 Telephone Telefax met@met.no Internet: met.no Bank account Swift code DNBANOKK

4 Postal address P.O.Box 43, Blindern NO-0313 OSLO Norway Office Niels Henrik Abelsvei 40 Telephone Telefax Internet: met.no Bank account Swift code DNBANOKK

5 Contents 1. Introduction Observations and data Temperature Precipitation Snow cover Homogeneity Methods Temporal variability of air temperature Temporal variability of precipitation Temporal variability of snow depth and snow-cover duration Summary and conclusions Acknowledgements References 32 5

6 1. Introduction The Arctic land areas have over the latest 2 3 decades experienced more warming than any other region on earth, and the sea-ice cover has decreased in the order of 10% in the same period (ACIA, 2005; IPCC, 2007). The Arctic climate conditions show large variability, both from year-to-year, but also on a decadal scale. A warm period, almost as warm as the present, was observed in the Arctic from 1925 to 1945, but its geographical distribution appears to have been different from the recent warming since the extent was not global (IPCC, 2007). IPCC (2007) states that most of the observed increase in globally-averaged temperatures since the mid-20th century is very likely due to the observed increase in anthropogenic greenhouse gas concentrations, and that it is likely that there has been significant anthropogenic warming over the past 50 years averaged over each continent except Antarctica. Climate models furthermore indicate that anthropogenic global warming will be enhanced in the northern high latitudes due to complex feedback mechanisms in the atmosphere ocean ice system. The climate changes seen in the Arctic have already led to major impacts on the environment and on economic activities. If the present warming continues as projected, these impacts are likely to increase, greatly affecting ecosystems, cultures, lifestyles and economies across the Arctic. The Arctic climate is a complex system and has multiple interactions with the global climate system. Changes in the Arctic climate are thus very likely to have significant impacts on the global climate system. These changes considerably influence the living conditions of the indigenous people in the northernmost regions and for reindeer husbandry in particular. While looking into the issues of the native population vulnerability and adaptation due to the climate change, the traditional reindeer herders experience is going to be used as well as the international climatic observation data network. (Tyler et. al. 2007). To estimate the vulnerability and the possibilities of the native population s adaptation to the climate change, we need looking into the temporal variability of the relevant climate parameters. Extreme climatic conditions are of specific interest. Other important issues for the vulnerability and community adaptation research are the projections of possible climate changes using climate models, and to describe the distribution of the meteorological parameters on a local scale. 6

7 The objective of this report is to describe the metadata and meteorological observation data made available from AARI for the EALAT and CAVIAR projects, and to outline the temporal tendencies of the climatic parameters which are most important for indigenous people in northern Russia, and particularly for reindeer husbandry. 2. Observations and data To analyze the long-term climate variability in the main reindeer husbandry areas of Russia (see fig.1); a dataset was created, containing the observations of air temperature, precipitation and snow depth. The dataset covers the whole period of observations from the end of the 19 th century (or the station starting point) up to 2008, and the temporal resolution was 1 day or 1 month. As outlined below, there were several changes and shifts in the observation system of observations. 2.1 Temperature The types of thermometers in use at each station remained the same throughout the period of record (Table 1). Minimum temperature was consistently measured with an alcohol thermometer, whereas hourly and maximum temperatures were measured by separate mercury thermometers. When the air temperature approached the freezing point of mercury (-38.9 C), either an alcohol thermometer, or in some cases a minimum thermometer alcohol column, was used in place of the mercury thermometer. The type of shelter or screen surrounding the thermometers varied considerably before In 1912, official instructions recommended sheltering thermometers with the Stevenson-type screen (before 1912, no such guidelines existed). However, it is likely that at many stations Stevenson screens were not implemented. From , Stevenson screens were replaced with the current screens at all stations. In 1928, additional guidelines regarding the exact dimensions of the shelters and their heights above the ground were issued (before 1928, no such specifications had been defined). Therefore, from 1930 on, most stations had their thermometers sheltered in roughly the same fashion. Major changes in the time of observation occurred in 1936 and Prior to 1936 measurements for computing daily mean temperature were taken at 07, 13, and 21 Local Mean Time (LMT) (minimum and maximum thermometers were checked at one of these 7

8 hours or at 09 LMT, see Table 1). Because of the lack of nighttime observations, daily mean temperature was probably overestimated by some location-dependent amount during this period. Beginning in 1936, all thermometers (hourly, minimum, and maximum) were checked at 01, 07, 13, and 19 LMT at most stations. As a result, the bias in daily mean temperature dropped to ~0.2 ºC. From 1966 to present, all thermometers were checked at 3-h intervals beginning at midnight Moscow winter Legal Time (MLT) (MLT being three hours later than Greenwich Mean Time). This rendered the bias in daily mean temperature insignificant. Table 1. Temperature recording methods and instrumentation Year Recording method/instrumentation implemented Measurements for computing daily mean temperature taken at 07, 13, and 21 LMT; mercury thermometer used; because of lack of nighttime observations, daily mean temperature probably overestimated 1881 Daily minimum temperature thermometer checked at 09 LMT; alcohol thermometer used Daily maximum temperature thermometer checked at 09 LMT; mercury thermometer used No regulations regarding type of shelter surrounding thermometers Daily minimum temperature thermometer checked at 07 and 21 LMT (lower value chosen); multiple measurements taken only to determine approximate time of occurrence of minimum Daily maximum temperature thermometer checked at 13 and 21 LMT (higher value chosen); multiple measurements taken only to determine approximate time of occurrence of maximum Official meteorological instructions recommended use of Stevenson screen to shelter thermometers; practice not implemented at all stations Official meteorological instructions recommended use of current screen to shelter thermometers; practice implemented over next ten years Official meteorological instructions specified exact size/height of screens Measurements for computing daily mean temperature taken at 01, 07, 13 and 19 LMT (or at 07, 13, 19, and 21 LMT); bias in daily mean temperature dropped to ~0.2 C; daily maximum and minimum thermometers may or may not have been checked each hour Measurements for all temperature variables collected at 3-h intervals beginning at midnight MLT; bias in daily mean temperature eliminated. 8

9 2.2 Precipitation The type of rain gauge used at each station changed at least once during the period of record (Table 2). In particular, the old-style gauge (type unknown) was replaced with the Tretyakov-type gauge over the period Whether or not other gauge replacements occurred at each station is not currently known. The type of shielding surrounding the rain gauges varied considerably over time. For example, in 1883, official instructions recommended that cross-shaped zinc strips be inserted into the gauge to prevent snow from drifting. Other shielding guidelines were issued at various times over the next half-century, up until the Tretyakov-type gauge was introduced. However, exact information on wind shields at each station is not currently known. Changes in the time of observation occurred in 1936, 1966, and 1986 (see Table 2). Before 1936, rainfall was measured only at 07 LMT. From , gauges were checked at 07 and 19 LMT. Beginning in 1966, the time of observation became time-zone dependent (the USSR comprised 11 time zones). Table 2. Precipitation recording methods and instrumentation Year Recording method/instrumentation implemented Rain gauge measurements taken at 07 LMT; snowfall converted to a liquid total by melting snow in gauge; type of gauge and shielding not standardized Official meteorological instructions recommended that cross-shaped zinc strips be inserted into the gauge to prevent snow from blowing out of the gauge; change probably not implemented at all stations Official meteorological instructions recommended surrounding the gauge with the funnel-shaped Nifer's shield; change probably not implemented at all stations Official meteorological instructions recommended erecting a fence around the gauge; change probably not implemented at all stations Official meteorological instructions recommended erecting a double fence around the gauge; change probably not implemented at all stations Rainfall measurements taken at 07 and 19 LMT; daily total rainfall obtained by summing all measurements for the calendar day Old-style gauge replaced with the Tretyakov-type gauge Rainfall measurements taken at 03, 09, 15 and 21MLT in time zone 2; at 03, 06, 15 and 18 MLT in zones 3-5; at 03 and 15 MLT in zones 6-8; at midnight, 03, 12, and 15 MLT in zones 9-11; and at 21, 03, 09 and 15 MLT in zone 12; wetting corrections <=0.2 mm applied to each hourly measurement.(because 4 observations per day were collected at stations in time zones 2-5 and 9-12, four corrections were counted in the daily total; therefore, total daily corrections are higher for stations in these areas.) 1986 Rain gauge measurements at 03 and 15 MLT discontinued at all stations except those in time zone 2. 9

10 2.3 Snow cover Snow depth measurement at the meteorological stations was carried out once a day. The value of the snow depth is assessed as an arithmetic average of measurements from three snow-measuring rods, which are usually placed on the meteorological measurement site. The measurements are influenced by the site s sheltering from the wind. Snow cover on the ground is estimated visually. Definition of a day with snow cover is when more than a half of the visible surface is covered with snow, regardless of its depth. The snow cover is considered as stable if it is observed continuously all winter or with no more than 3 days gaps in each 30 days of its observation. It is also assumed that a daily gap in the beginning of the winter is to be preceded by at least 5 days with snow covering, while 2 or 3 days gaps by at least 10 such days. 2.4 Homogeneity A relocation of a station, as well as a change in the observation procedures or in the gauges type can lead to inhomogeneities in the climate series. Main causes of inhomogeneities are: tendencies of station relocation to more open space in the 1930s. In 1936 a shift in the number of daily temperature measurements (4 times instead of 3 before) revealed that the mean daily values were overestimated before the shift. The change to Tretyakov precipitation gauge in the 1950s led to increased catch efficiency; especially for solid precipitation. The homogeneity of the snow depth series is influenced by relocations of the rods to sites with different wind sheltering. The influence of possible inhomogeneities should be considered when analyzing the long-term variability. Table 3. Station relocations and rain gauge replacement dates. Station name WMO number Station relocation Rain gauge replacement Lovozero Sojna Ust Cilma ,1904 -, , , , , Salekhard Mare-Sale Nadym Tarko-Sale

11 Kusur Djalinda Gigansk , , , Djardan Ust-Moma Olenek Tompo Oimakon Yakutsk Markovo , , , Anadyr , Metadata for the stations are presented in Table 4. Aria Table 4. Survey of meteorological stations used in this report WMO number Station name Distance (km) to nearest settlement Latitude, degrees Longitu de, degrees Elevation (m a.s.l.) Start (year) Kola Lovozero peninsula Nenets Sojna AA Republic Ust Cilma Komi Yamalo Tarko Nenets AA Sale Salekhard Mare-Sale Nadym Republic Kusur Sakha Gigansk Djardan Djalinda Ust-Moma Yakutsk Olenek Tompo Oimakon Chukci Markovo AA Anadyr Location of the stations is shown in Figure 1. 11

12 Figure 1. Meteorological stations location. Numbers on the map are WMO numbers The meteorological observations data were taken from the following sources: USSR climate handbooks, Meteorological monthly and annual summaries released by the Main Geophysical observatory, All-Russian hydrometeorologial University, Arctic and Antarctic Research Institute. We also used the information from the following websites: All-Russian Research Institute website, The Royal Netherlands Meteorological Institute, Carbon Dioxide Information Analysis Center (USA), National Climate Data Center (USA), National Snow and Ice Data Center (USA) websites. Partly the information was received directly from the Murmansk and Tiksi meteorological services administration, as well as from the Arctic and Antarctic Research Institute archives. Table 5 gives an overview of observation periods for each parameter included in the dataset. 12

13 Table 5. Available climate series for the analysis. Station Temperature Precipitation Snowdepth Name WMO monthly daily monthly daily monthly period daily Lovozero Soina Ust-Cilma Tarko-Sale Salekhard Mare-Sale Nadym Kusur Gigansk Djardjan Djalinda Ust-Mona Yakytsk (1830) Olenek Tompo Oimakon Markovo Anadyr Methods The statistical methods used in this study were focusing on the long-term variability and possible climate regime shifts assessment. Non-parametrical approaches which do not require a priori restrictions for the analyzed series were preferred, as parametrical methods require a specific distribution type of the series. Due to lack of data it was difficult to determine whether the distribution type of the series obey the parametric approach conditions. To test the statistical hypothesis of an existing monotonous trend the Mann-Kendall test was used (Mann 1945, Kendall 1975). The value of a trend s bias, assuming its linearity, was determined using the Sen s slope test (Gilbert 1987, Sirois 1998). The year of trend shifts were assessed using the Pettitt test (Pettitt 1979), as well as the wavelet analysis. A continuous expansion by the orthogonal Daubechis 4 th degree wavelet was used (Solonia & Arbuzov, 2008). 13

14 In addition, mean value series shift detection was carried out using the Rodionov tool (Rodionov 2004), which includes bilateral T-criteria and Huber weight parameter to estimate the deviation between the mean values. The moving averages were assessed by 10 values with the temporal shift equal to 1. What made the analysis complicated was the large number of gaps in the observation data. To make the results for each element comparable, an observation period with no gaps was chosen for most stations. 4. Temporal variability of air temperature Seasonal and annual temperature averages were calculated using the mean monthly data: March - May for spring, June August for summer, September November for autumn and December (of the previous year) February for winter. The calculated averages were analyzed for the period The estimates of years with a tendency shift using Petitt test are seen in Table 6. A tendency of decreasing temperature in the first part of the series was replaced by an increasing tendency in the later decades. According to all studied stations data, a winter shift occurred earlier than a summer shift. Meanwhile, when looking into the spring and fall data, it is clear that at the stations west of Yakutsk the shift was observed earlier than at the stations east of Yakutsk. Table 6. Temperature change point detection (based on Pettitt s test) station Annual Spring Summer Autumn Winter Lovozero Sojna Ust-Cilma Tarko-Sale Salekhard Mare-Sale Kusur Gigansk Djardan Ust-Moma Yakutsk Olenek Tompo Oimakon Markovo Anadyr

15 Estimates of linear temperature trends are shown in Table 7. Statistically significant monotonous trends were defined using Mann-Kendall criteria with the significance level equal to It is clear from Table 7 that there are no statistically significant trends in the annual average series for Kola Peninsula, Nenets AA and Yamalo-Nenets AA, Komi republic, while there are significant trends in the data for Sakha republic and Chukci AA. The same tendency is seen for the seasonal data as well, significant trends for winter data are found only for Sakha. Remarkable are the negative trends for autumn and winter, indicating decreasing air temperatures, although these trends are not statistically significant. However, the fact that the stations with negative trends are situated close to each other gives a reason to assume it is not a coincidence. Table 7. Linear temperature trends (degc/year) during (bold indicate trends significant at the 0.05 level) station Annual Spring Summer Autumn Winter Lovozero Sojna Ust-Cilma Tarko-Sale Salekhard Mare-Sale Kusur Gigansk Djardan Ust-Moma Yakutsk Olenek Tompo Oimakon Markovo Anadyr With the use of the Rodionov tool, the climate regime shift dates were investigated. Each climate regime was characterized by the long-term annual temperature averages for a specific period of time, where temperature values were averaged seasonally or annually. The shifts dates are shown in Table 8, as well as the deviations between the mean values, considering Δ = T k Tk 1. The calculations were provided for each station for different periods depending on the duration of observation and the number of gaps. 15

16 Table 8. Change dates and differences ( C ) in shifts in average values of air temperature Lovozero: (Obs. period ). Winter: 1989/+1.5, spring: 2003/+1.6, Summer: 1975/- 0.8 and 1999/+1.6, annual: 2003/+1.5 Sojna: ( ). Winter: 1965/-1.3, summer: 2004/+1.5, autumn: 1964/-0.9 and 2003/+1.8 Ust Cilma: ( ). Spring: 1989/+1.5, summer: 2003/+1.3, autumn: 1952/-1.2 and 2003/+2.5, annual: 1963/-1.2, 1972/+1.2 and 2003/+1.7 Tarko Sale: ( ). Spring: 1957/-1.5 and 1990/+2.6, summer: 2000/+1.7, autumn: 1957/ -1.6 and 1977/+1.4, annual: 1957/-1.2 and 1980/+1.5 Salekhard: ( ). Spring: 1943/+5.2, 1957/-2.2, 1989/+2.9 and 2004/-4.3, summer: 1904/+1.3, 1968/-1.3 and 1981/+1.8, autumn: 1895/+2.4 and1964/-0.7, annual: 1903/+1.6, 1963/-1.3 and 1981/+1.5 Mare-Sale: ( ). Spring: 1943/+2.3, 1957/-2.0, 1989/+3.0 and 1998/-3.0, summer: 1953/+1.5, 1963/-1.5 and 1989/+1.3, autumn: 1956/-2.0 and 2004/+3.0, annual: 1963/-1.4, 1981/+1.1 and 2005/+2.0 Kusur: ( ). Winter: 1941/-2.2, spring: 1955/-1.4 and 1971/+1.5, summer: 2005/+1.3 Autumn: 1953/-1.7 and 2001/+1.8, annual: 1958/-0.7 and1993/+1.0 Gigansk: ( ). Winter: 1989/+1.5, spring: 2000/+1.8, autumn: 1952/-1.2, annual: 1950/ -0.6 and 1988/+0.9 Djardan: ( ). Winter: 1989/+1.3, spring: 1955/-1.4 and 1971/+1.4, summer: 1997/+1.0, autumn: 1952/-2.0, 2005/+1.6, annual: 1950/-1.0, 1988/+0.7 and 2005/1.0. Ust Moma: ( ). Winter: 1976/+2.4, spring: 2000/+2.4, summer: 1997/+1.2, autumn: 1994/+2.2, annual: 1985/+1.5 and 2005/+1.4 Yakutsk: ( ). Winter: 1990/+3.1, spring: 1895/+0.7, 1974/+1.5 and 2002/+2.1, Summer: 1893/+0.7, autumn: 2005/2.2, annual: 1988/1.5 Olenek: ( ). Winter: 1967/+3.5, spring: 1955/-2.1 and 1965/+2.7, Annual: 1950/- 1.4, 1967/+2.0 and 2005/+1.7 Tompo: ( ). Spring: 1954/-1.3 and 1968/+2.2, Summer: 1997/+1.0, Autumn: 2005/+2.4, Annual: 1980/+0.9 Oimakon: ( ). Spring: 1954/-1.7 and 1966/+2.1, Summer: 1995/+1.1, Autumn: 1955/ -0.8 and 2002/+2.5, Annual: 1980/+0.9 and 2005/+1.0 Markovo: ( ). Winter: 1978/+2.2 and 1989/-2.8, Spring: 1935/+2.7, 1944/-1.8, 1996/+3.0 and 2005/-2.3, Summer: 1920/+1.2, 1946/-1.8, 1953/+1.5 and 2002/+1.5, Autumn: 1921/+1.4, 1954/-2.1, 1970/+0.8 and 1995/+2.2, Annual: 1921/+1.5, 1946/-0.6 and1995/+1.0 Anadyr: ( ). Winter: 1978/+2.0, 1989/-3.0 and 2004/+4.0, Spring: 1935/+2.6, 1944/ -1.8, 1996/+3.0 and 2005/-2.1, Summer: 1920/+1.2, 1946/-1.8, 1953/+1.6 and 2005/-1.6, Autumn: 1921/+1.4, 1954/-2.1, 1970/+0.8, 1995/+2.2 and 2005/+3.5, Annual: 1921/+1.5, 1946/ -0.6 and 1995/+1.2 Table 8 reveal the repeated shifts (positive and negative); both in the annual and seasonal averages. The changes in the mean mainly correspond the mean temperature for the whole hemisphere, although the range of the change points is rather wide. Most stations see a significant warming in the late 1990s 2000s. Some peculiarities are remarkable. For the stations situated in Nenets AA and Yamalo-Nenets AA and Komi republic no shift in the mean is found in winter. In 1965 Sojna had a negative change of С in annual temperature. 16

17 Also in Yamalo-Nenets AA negative temperature changes are found in spring (-4 0 С for Salekhard in 2004 and -3 0 С for Mare-Sale in 1998), although there is an obvious warming for the whole year. Also in Chukci AA negative changes are found during the spring season: - 2,1 0 С for Anadyr in 2005 and С for Markovo in An additional analysis of the change tendencies using the same tests was carried out on the number of days when the daily mean daily air temperature was above/below certain thresholds. Because of missing data, it was only possible to perform this analysis for 6 stations covering the period from 1943 to Based on information from indigenous people, number of days with daily mean temperature below С, above 0 0 С and above С is especially important for the native population s living conditions. Trends in frequencies of temperatures above /below these limits are presented in Table 9. The negative trend in the number of the days with extremely low temperatures (corresponding to the observed warming) is observed in the data for Yakutsk;- the rest of the stations have no such a trend. Chukci district had an increase in number of days with temperatures above С as well as above 0 0 С, which confirms the assumption of a summer warming in this area. Kola Peninsula also has an increase in number of days with mean temperatures above 0 0 С. Table 9. Linear trends (days/year) during in number of days with temperatures above/below given thresholds (bold indicate trends statistically significant at the 0.05 level) Station T<-30 C T>10 C T>0 C Lovozero Ust Cilma Gigansk Yakutsk Markovo Anadyr Temporal variability of precipitation Changes in precipitation amount were studied in the same way as for air temperature. The analysis was carried out for the stations with no more than one gap for the period The results of the change points study are shown in the Table 10. In comparison with the temperature data, the change points for precipitation data are observed earlier on most stations, mainly s. Such an outcome in some cases might be caused by the change in gauge types (see Table 3) 17

18 Table 10. Precipitation change points detection Station Annual Spring Summer Autumn Winter Lovozero Sojna Ust-Cilma Tarko-Sale Salekhard Mare-Sale Kusur Gigansk 1956 Djardan Yakutsk Olenek Markovo Anadyr 1955 Table 11 shows that significant positive trends in precipitation are revealed in all seasons but summer;- the strongest trends are found in spring and winter. For all stations except Yakutsk (negative trend for winter season) the precipitation amount is increasing. No significant trends are revealed for Ust-Cilma, Salekhard and Sojna. The strongest significant trends are found for Lovozero in winter and for the whole year, at Mare-Sale for the whole year, and at Anadyr in winter. The analysis reveals the strong influence of inhomogeneities in precipitation series caused by the introduction of the Tretyakov gauge (cf. Table 3). Table 11. Linear trends in precipitation amount (mm/year) during (bold indicate trends statistically significant at the 0.05 level) Station Annual Spring Summer Autumn Winter Lovozero Sojna Ust-Cilma Tarko-Sale Salekhard Mare-Sale Kusur Gigansk Djardan Ust-Moma Yakutsk Olenek Tompo Oimakon Markovo Anadyr

19 Figures 2-13 illustrate the temporal variability of the annual precipitation sums for the period from s up to 2008, as well as the spectrograms of the first coefficient of the continuous wavelet expansion with the use of Daubechis 4 degree wavelet. The values of the coefficient are depicted by its absolutes, with colour scaling. The light stripes on the spectrograms correspond the moments of abrupt precipitation sum changes. Fig. 4 and 13 depict the winter and annual precipitation tendencies at Mare-Sale station. It is clear from the figures that there is an inhomogeneity in the temporal series caused by the new rain gauge introduction in Figure 2. Temporal variability and spectrogram of the 1 st coefficient of annual precipitation sum at Lovozero station. 19

20 Figure 3. Temporal variability and spectrogram of the 1 st coefficient of annual precipitation sum at Sojna station. Figure 4. Temporal variability and spectrogram of the 1 st coefficient of annual precipitation sum at Mare-Sale station. 20

21 Figure 5. Temporal variability and spectrogram of the 1 st coefficient of annual precipitation sum at Salekhard station. Figure 6. Temporal variability and spectrogram of the 1 st coefficient of annual precipitation sum at Ust Cilma station. 21

22 Figure 7. Temporal variability and spectrogram of the 1 st coefficient of annual precipitation sum at Nadym station. Figure 8. Temporal variability and spectrogram of the 1 st coefficient of annual precipitation sum at Tarko-Sale station. 22

23 Figure 9. Temporal variability and spectrogram of the 1 st coefficient of annual precipitation sum at Ust-Moma station. Figure 10. Temporal variability and spectrogram of the 1 st coefficient of annual precipitation sum at Yakutsk station. 23

24 Figure 11. Temporal variability and spectrogram of the 1 st coefficient of annual precipitation sum at Markovo station. Figure 12. Temporal variability and spectrogram of the 1 st coefficient of annual precipitation sum at Anadyr station. 24

25 Figure 13. Temporal variability and spectrogram of the 1 st coefficient of precipitation sum in winter at Mare-Sale station. The analysis of time changes of precipitation has been performed by means of comparison of averages by the Rodionov s method. Results are presented in Table 12. The analysis is performed for the longest possible period for each station. On the majority of the 11 stations numerous changes of average values are found; - both positive and negative. 25

26 Table 12. Change dates and differences (mm) in shifts in average values of precipitation sums Lovozero ( ), Winter: 1952/+37, Spring: 1948/+23 and 1966/+20, Autumn: 1962/+28, 1986/-18 and 2004/+27, Annual: 1948/+88 and 1961/+76 Sojna ( ) No changes Ust Cilma ( ), Winter:1944/+31 and 1965/-8, Spring: 1943/+14, Autumn:1938/+40 and 1967/-30, Annual:1940/+125, 1952/-67 and 2006/+135 Tarko-sale ( ), Winter: 1974/+19, Spring: 1985/+31, Summer:1957/-60 and 1997/+70, Autumn: 2004/+30, Annual:1959/+47 and 1991/+77 Salekhard ( ), Winter: 1951/+26 and 1964/-15, Autumn: 1951/+63 and1960/-50, Mare-Sale ( ), Winter: 1953/+39, 1972/+8 and 2001/-34, Spring: 1950/+23, 1965/+14 and 1998/-14, Autumn: 1951/+28 and 2003/+25, Annual: 1951/+103 and 2002/-50 Kusur ( ), Summer: 1988/+35, Autumn:1962/+37, 1974/-96 and 1985/+21, Annual: 1988/+87, 1995/-110 and 2002/-115 Gigansk ( ), Winter:1976/+10 and 2002/+19, Autumn: 1992/+26, Annual: 1978/ /-85 and 2005/+45 Djardan ( ), Winter: 2002/+10, Autumn: 1978/-25 and 2000/+28, Annual:1988/+78 Ust Moma ( ), Winter: 1966/+4 and 1987/-9, Summer: 1970/+23, Autumn: 2004/+18 Yakutsk ( ), Winter: 1954/+8, Spring:1966/+18 and 1973/-15, Summer:1974/+30, 1985/-32 and 2005/+45, Autumn: 2006/+18, Annual: 1974/+27 and 2005/+46 Olenek ( ), Spring: 1988/+14 and1995/-10 Tompo ( ), Winter: 1991/+5 Oimakon ( ), Winter:1981/-9, Spring: 1971/-5, Summer: 1975/+15, Autumn:2006/+16 Markovo ( ), Winter: 1956/+53 and 1973/-26, Spring:1954/+22, Summer: 2000/- 39, Annual: 1959/+77 Anadyr ( ), Winter: 1965/+66, 1971/-85, 1979/+104, 1990/-73 and 2000/+49, Autumn: 1997/+45 and 2006/-56, Annual: 1997/+88 Figures reveal that inhomogeneities are found in precipitation series from Mare-Sale in all seasons except summer. 26

27 Shifts in the mean for wi, Probability = 0.1, cutoff length = 10, Huber parameter = Figure 14. Shifts in mean precipitation sums at Mare-Sale in winter for the period Shifts in the mean for sp, Probability = 0.1, cutoff length = 10, Huber parameter = Figure 15. Shifts in mean precipitation sums at Mare-Sale in spring for the period

28 Shifts in the mean for su, Probability = 0.1, cutoff length = 10, Huber parameter = Figure 16. Shifts in mean precipitation sums at Mare-Sale in summer for the period Shifts in the mean for au, Probability = 0.1, cutoff length = 10, Huber parameter = Figure 17. Shifts in mean precipitation sums at Mare-Sale in autumn for the period

29 Shifts in the mean for rg, Probability = 0.1, cutoff length = 10, Huber parameter = Figure 18. Shifts in mean precipitation sums at Mare-Sale for annual values during the period The analysis of the precipitation intensity is provided for the period The change in annual maximum of daily precipitation was calculated, as well as intensive (over 10mm/day) and moderate (over 1mm/day) precipitation frequency. The trends of these parameters are outlined in Table 13. Table 13. Linear trends ( ) in maximum 1 day precipitation (mm/year) and number of days with precipitation >1 & 10 mm (days/year). Bold indicate trends statistical significant at the 0.05 level. Station Daily max R>10 mm R>1 mm Lovozero Ust Cilma Tarko-Sale Salekhard Mare-Sale Kusur Gigansk Yakutsk Oimakon Markovo Anadyr

30 An increasing trend in daily precipitation maximums is observed only for Anadyr, while increase in precipitation frequency is found for Anadyr and Lovozero. Almost all stations except Oimakon and Markovo have an increase in moderate precipitation frequency. The most significant increase is observed at Kusur, Anadyr, Lovozero and Mare-Sale stations, which is clear from the trend biases analysis. 6. Temporal variability of snow depth and snow cover duration In this section we are looking into the changes of the snow cover characteristics, which are important for the reindeer breeding. Most analyzed stations experience the highest value of snow depth in April. Linear trends for snow depth for April and in duration of snow cover (from autumn till spring) are presented in Table 14 and 15. Table 14. Linear trends and change points (Pettitt) in duration of stable snow cover ( ). Bold indicate trends statistical significant at the 0.05 level. Station Trend slope Pettitt test Kusur Mare-Sale Salekhard Ust Cilma Gigansk Oimakon Yakutsk Markovo Anadyr Table 15. Linear trends and change points (Pettitt) in snow depth for April ( ). Bold indicate trends statistical significant at the 0.05 level. Station Trend slope Pettitt test Lovozero Tarko-Sale Ust Cilma Kusur Gigansk Oimakon Yakutsk Markovo Anadyr

31 According to the outcome depicted in Table 14, significant trends are observed only for 2 stations Mare-Sale and Gigansk, and the trend is positive, which indicates an increase of the continuous snow cover duration. Data from Table 15 indicate increase in snow depth for 4 stations. The strongest increase is observed at Tarko-Sale and Ust Cilma. The increase in snow duration (Table 14) and snow depths (Table 15) despite increasing temperatures (Table 7) is probably due to increased precipitation during the cold season (Table 11). 7. Summary and conclusions Long-term variability in temperature, precipitation and snow conditions have been studied for 18 localities in the Kola Peninsula, Nenets AA, Komi Republic, Yamalo-Nenets AA, Sakha Republic and Chukci AA in northern Russia. An overview is given of station metadata (length of series, relocation of stations, observing hours, instrumentation etc.) and of methodology to study long-term trends and shifts in various climate series. For temperature no statistically significant trends ( ) were found in the annual average series for Kola Peninsula, Nenets AA and Yamalo-Nenets AA, Komi republic, while there are significant positive trends in the data for Sakha republic and Chukci AA. The same tendency is seen for the seasonal data as well, significant trends for winter data are found only for Sakha. Some negative trends are found for autumn and winter, but these trends are not statistically significant. At most stations there are a significant warming after A negative trend ( ) was found in number of days with very low temperatures (daily mean temperature below -30 C). In the Chukci district there was an increase in number of days with mean temperature above 0 and +10 C. Also for the Kola Peninsula there has been an increase in number of days with mean temperatures > 0 C. For precipitation significant positive trends ( ) were found for all seasons except summer, - with strongest trends during winter and spring. The strongest trends were found for the Lovozero station at the Kola Peninsula. The precipitation trends are however influenced by inhomogeneities caused mainly by changes in precipitation gauges during the period For duration of snow cover two stations (Mare-Sale and Gigansk) had significant trends. At both stations the trends were positive, i.e. increased length of the period with continuous snow cover. The highest snow depths in northern Russia are usually observed in 31

32 April. For 4 stations significant positive trends were found in snow depths for April;- the strongest increase is observed at Tarko-Sale and Ust Cilma. The reason for the increase in snow duration and maximum snow depths despite increasing temperatures is probably increased amounts of solid precipitation during the cold season. 8. Acknowledgements This work is supported financially by The Research Council of Norway through two IPYprojects (EALAT-RESEARCH and CAVIAR) and by The Nordic Council of Ministers. We particularly want to thank Svein Disch-Mathiesen at the Saami University College in Kautokeino and Grete Hovelsrud at the CICERO Institute in Oslo for promoting and supporting the collaboration between AARI and met.no. Warm thanks also to Øyvind Nordli, met.no for scrutinizing and commenting the manuscript. 9. References ACIA Arctic Climate Impact Assessment Scientific Report. Cambridge University Press, Cambridge, , ISBN: Gilbert R.O Statistical methods for environmental pollution monitoring. Van Nostrand Reinhold, New York. IPCC, 2007: Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change (S. Solomon, D. Quin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt, M. Tignor and H.L. Miller (ed.)). Cambridge University Press Mann H.B Nonparametric tests against trend. Econometrica, 13, Kendall M.G Rank Correlation Methods, Griffin, London. Pettitt A.N A nonparametric approach to the change-point problem. Applied Statistics, 28, Rodionov S A seqential algorithm for testing climate regime shifts. Geophysical Research Letters, 31, -L Sirois, A A Brief and Biased Overview of Time Series Analysis or How to Find that Evasive Trend. In WMO report 133: WMO/EMEP Workshop on Advanced Statistical Methods and their Application to Air Quality Data sets (Helsinki, 14-18, September, 1998). Solonia, A.I., & S.M.Arbuzov, 2008: Digital processing of signals. Modelling in MatLab. St.Petersburg, Russia Tyler N.J.C., Turi J.M., Sundset M.A., Strom Bull K., Sara M.N., Reinert E., Oskal N., Nellemann C., McCarthy J.J., Mathiesen S.D., Martello M.L., Magga O.H., Hovelsrud G.K., Hanssen-Bauer I., Eira N.I., Eira I.M.G., Corell R.W., 2007: Saami reindeer pastoralism under climate change: Applying a generalized framework for vulnerability studies to a sub-arctic social-ecological system. Global Environmental Change, 17, , 32